# library(sjlabelled) #package to remove data labels. Labels slow down analysis
# library(tidyverse)
# library(kableExtra)
# library(formattable)
#
# d <- d %>%
# remove_all_labels() %>% filter(EMPSTAT != 1) %>% #excluding military personnel
# select(-CPSID,-CPSIDP) %>% mutate(
# famid = SERIAL + MONTH / 10,
# #family is is unique for each family record.
# id = SERIAL + MONTH / 10 + PERNUM / 1000,
# #id is unique for each individual record
# sex = ifelse(SEX == 1, "Men", "Women"),
# married_spouse_present = (MARST == 1) * 1,
# married = (MARST %in% c(1, 2)) * 1,
# unemployed = (EMPSTAT %in% c(20, 21, 22)) * 1,
# employed = (EMPSTAT %in% c(1, 10, 12)),
# retired = (EMPSTAT == 36) * 1,
# lfp = (LABFORCE == 2) * 1,
# age16plus = (AGE > 15) * 1,
# workingage = (AGE < 63 &
# AGE > 17) * 1,
# year_month = YEAR * 100 + MONTH,
# period_num = (YEAR %% 2020) * 12 + MONTH
# ) %>% select(YEAR,
# MONTH,
# STATEFIP,
# WTFINL,
# age16plus,
# workingage,
# sex,
# lfp,
# unemployed,
# period_num)The COVID pandemic has led to unemployment spikes and prolonged high unemployment throughout the U.S. Some states have been impacted more than others. The states that stand out in their adverse reaction to the limitations and behavioral changes due to the pandemic are the states heavily dependent on tourism and entertainment, in particular Nevada and Hawaii. During the month of peak unemployment, in April 2020, Nevada had the highest unemployment rates among all U.S. states, at 30%, Michigan had the next highest unemployment rate, at 24%, and in Hawaii unemployment reached 22%. In February 2021 unemployment rate in Michigan was close to its pre-pandemic level, at 5%. Nevada and Hawaii were still among the 10 states with highest unemployment rates in February 2021, at 8-9%. Before the pandemic unemployment rates in these states were 3-4%.
economic impact of the pandemic in the U.S. varied greatly depending on the state of residence. I look at U.S. unemployment dynamics starting in January 2020 focusing on regional differences in the size of the initial shock to employment and the speed of recovery. The data come from the Current Population Survey (Flood et al. 2021) — a monthly survey administered by the Census Bureau and used to calculate the official unemployment statistics.
Non-coastal Northwestern states with low population densities such as North and South Dakota, Nebraska, and Wyoming had unemployment rates below 10% during peak unemployment months, and returned to 3-6% unemployment rates in December. California, New York, and New Jersey with high population densities and large urban centers had unemployment rates above 10% throughout most of 2020 and 7-9% unemployment rates in December.
Some of the states impacted the most were the states with a high share of workers in entertainment, tourism and food service. Nevada, with 16% of GDP from tourism, had the highest peak in unemployment, 30% in April 2020. The industries particularly hit by COVID in Nevada were food and beverage and tourism (2020). Hawaii had an unemployment rate of 22% in April, and has been the slowest state to recover, with unemployment rate at 10% in December 2020.
In most states unemployment peaked in April or May and declined throughout the rest of the year, until October or November. December saw an increase in unemployment rates in many states including California, New Mexico, Illinois, Rhode Island, Michigan, Colorado and Washington.
# library(srvyr)
#
# tab <-
# d %>% as_survey(weights = c(WTFINL)) %>% #represent data as a survey with sampling weights
# filter(age16plus == 1, lfp == 1) %>% #exclude individuals under 16. Choose the labor force as the base for unemployment rates
# group_by(period_num, STATEFIP) %>% summarize(
# year = YEAR[1],
# month_num = MONTH[1],
# month = month.abb[MONTH[1]],
# period = paste(month.name[MONTH], YEAR)[1],
# unemployment_rate = survey_mean(unemployed)
# ) %>% #standard errors are the standard errors of the sample mean that account for the uncertainty of each observation (weights away from 1 mean that the observations deviation from the mean increases the uncertainty, increasing se).
# #The variance attains its maximum value, when all weights except one are zero. Its minimum value is found when all weights are equal (i.e., unweighted mean), in which case it degenerates into the standard error of the mean, squared.https://en.wikipedia.org/wiki/Weighted_arithmetic_mean
# ungroup()
#
# tab[, 7:8] <-
# lapply(tab[, 7:8], percent, 1)#format all columns except year_month as percentagesThe maps below show the changes in unemployment rates in the contiguous U.S. by state during 2020. The states in non-coastal Northwest had a smaller spike in unemployment and quicker recovery. States with large metropolitan centers, such as California, New York, and Illinois and states heavily depend on tourism, including Nevada and Hawaii, were still experiencing high unemployment rates, 8-10%, in December 2020.
# library(urbnmapr) #package to look up state name by FIPS
# library(maps)#package for creating state maps with ggplot
#
# states <- states %>% mutate(state_fips = as.integer(state_fips))
#
# state_lookup <-
# states %>% filter(!duplicated(states$state_fips)) %>% select(state_fips, state_name)
#
# tab <-
# full_join(tab, state_lookup, by = c("STATEFIP" = "state_fips")) %>%
# select(
# period_num,
# period,
# year,
# month_num,
# month,
# state_name,
# unemployment_rate,
# unemployment_rate_se
# )
#
# save(tab, file = "tab.RData")
#
# mapdata <- map_data("state") %>%
# mutate(region = str_to_title(region)) %>% mutate(state_name = str_replace(region, "District Of Columbia", "District of Columbia")) %>% select(long, lat, group, state_name)
#
# mapdata <-
# left_join(mapdata, tab, by = "state_name")
# save(mapdata, file = "mapdata.RData")library(tidyverse)
library(plotly)
library(ggplot2)
library(ggthemes)
load("tab.RData")
load("mapdata.RData")
mid <-
mean(tab[!(tab$state_name %in% c("Alaska","Hawaii")),]$unemployment_rate) #unemployment midpoint used to determine the color scheme
period_names <- unique(tab$period)
period_labeller <- function(variable, value) {
return(period_names[value])
}
p <- ggplot(data = mapdata,
aes(
x = long,
y = lat,
group = group,
fill = unemployment_rate,
text = state_name
)) +
facet_wrap(period_num ~ ., ncol = 2, labeller = period_labeller) +
geom_polygon(color = "gray50", size = 0.1) +
coord_map(projection = "albers",
lat0 = 39,
lat1 = 45) +
scale_fill_gradient2(
midpoint = mid,
low = "purple",
mid = "white",
high = "orangered",
space = "Lab",
labels = scales::percent_format()
) +
labs(title = "", fill = NULL) +
theme_map()
ggplotly(p, height = 800,
width = 800)Source: Current Population Survey: https://cps.ipums.org
The tables below shows unemployment rates in February (before the economic impact of the pandemic), April (peak unemployment month for most states) and in February 2021.
load("tab.RData")
library(DT)
tab <- tab %>% mutate(
year = as.integer(year),
month_num = as.integer(month_num),
unemployment_rate = round(as.numeric(unemployment_rate), 2),
"unemployment_rate_se" = round(as.numeric(unemployment_rate_se), 2)
) %>% select(-period_num,-period)
DT::datatable(
tab,
colnames = c(
"Year",
"Month Number",
"Month",
"State",
"Unemployment Rate",
"Standard Error"
),
filter = "top",
extensions = 'Buttons',
options = list(dom = 'Blfrtip',
buttons = c('copy', 'csv', 'excel'))
) %>% formatPercentage(c("unemployment_rate", "unemployment_rate_se"), 2)“Coronavirus Will Slam States Dependent on Tourism, Stateline Article.” 2020. https://www.pewtrusts.org/en/research-and-analysis/blogs/stateline/2020/03/16/coronavirus-will-slam-states-dependent-on-tourism.
Flood, Sarah, Miriam King, Renae Rodgers, Steven Ruggles, and J. Robert Warrenumley. 2021. “Integrated Public Use Microdata Series, Current Population Survey.” https://cps.ipums.org/cps.
“Nevada’s 28.2 Percent Unemployment Rate Eclipses Entire United States. It’s Worse Than the Great Depression.” 2020. https://www.rgj.com/story/news/2020/05/22/nevada-unemployment-rate-soars-28-2-state-down-244-800-jobs/5243949002/; Reno Gazette Journal.
Stolpovsky, Elena. 2021. “Labor Force Participation and Employment of Men and Women in the United States During the Covid-19 Pandemic.” https://rpubs.com/elenas70/labormarket2020part1.